Dataset Information

Latent classes for chemical mixtures analyses in epidemiology: an example using phthalate and phenol exposure biomarkers in pregnant women.

ABSTRACT: Latent class analysis (LCA), although minimally applied to the statistical analysis of mixtures, may serve as a useful tool for identifying individuals with shared real-life profiles of chemical exposures. Knowledge of these groupings and their risk of adverse outcomes has the potential to inform targeted public health prevention strategies. This example applies LCA to identify clusters of pregnant women from a case-control study within the LIFECODES birth cohort with shared exposure patterns across a panel of urinary phthalate metabolites and parabens, and to evaluate the association between cluster membership and urinary oxidative stress biomarkers. LCA identified individuals with: "low exposure," "low phthalates, high parabens," "high phthalates, low parabens," and "high exposure." Class membership was associated with several demographic characteristics. Compared with "low exposure," women classified as having "high exposure" had elevated urinary concentrations of the oxidative stress biomarkers 8-hydroxydeoxyguanosine (19% higher, 95% confidence interval [CI]?=?7, 32%) and 8-isoprostane (31% higher, 95% CI?=?-5, 64%). However, contrast examinations indicated that associations between oxidative stress biomarkers and "high exposure" were not statistically different from those with "high phthalates, low parabens" suggesting a minimal effect of higher paraben exposure in the presence of high phthalates. The presented example offers verification of latent class assignments through application to an additional data set as well as a comparison to another unsupervised clustering approach, k-means clustering. LCA may be more easily implemented, more consistent, and more able to provide interpretable output.

SUBMITTER: Carroll R

PROVIDER: S-EPMC6917962 | biostudies-literature | 2020 Jan

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Latent classes for chemical mixtures analyses in epidemiology: an example using phthalate and phenol exposure biomarkers in pregnant women.

Carroll Rachel R White Alexandra J AJ Keil Alexander P AP Meeker John D JD McElrath Thomas F TF Zhao Shanshan S Ferguson Kelly K KK

Journal of exposure science & environmental epidemiology 20191021 1

Latent class analysis (LCA), although minimally applied to the statistical analysis of mixtures, may serve as a useful tool for identifying individuals with shared real-life profiles of chemical exposures. Knowledge of these groupings and their risk of adverse outcomes has the potential to inform targeted public health prevention strategies. This example applies LCA to identify clusters of pregnant women from a case-control study within the LIFECODES birth cohort with shared exposure patterns ac ...[more]

PMID: 31636370

Similar Datasets

Project description:BackgroundParental preconception exposure to select phenols and phthalates was previously associated with increased risk of preterm birth in single chemical analyses. However, the joint effect of phenol and phthalate mixtures on preterm birth is unknown.MethodsWe included 384 female and 211 male (203 couples) participants seeking infertility treatment in the Environment and Reproductive Health (EARTH) Study who gave birth to 384 singleton infants between 2005 and 2018. Mean preconception urinary concentrations of bisphenol A (BPA), parabens, and eleven phthalate biomarkers, including di(2-ethylhexyl) phthalate (DEHP) metabolites, were examined. We used principal component analysis (PCA) with log-Poisson regression and Probit Bayesian Kernel Machine Regression (BKMR) with hierarchical variable selection to examine maternal and paternal phenol and phthalate mixtures in relation to preterm birth. Couple-based BKMR model was fit to assess couples' joint mixtures in relation to preterm birth.ResultsPCA identified the same four factors for maternal and paternal preconception mixtures. Each unit increase in PCA scores of maternal (adjusted Risk Ratio (aRR): 1.36, 95%CI: 1.00, 1.84) and paternal (aRR: 1.47, 95%CI: 0.90, 2.42) preconception DEHP-BPA factor was positively associated with preterm birth. Maternal and paternal BKMR models consistently presented the DEHP-BPA factor with the highest group Posterior Inclusion Probability (PIP). BKMR models further showed that maternal preconception BPA and mono(2-ethyl-5-hydroxyhexyl) phthalate, and paternal preconception mono(2-ethylhexyl) phthalate were positively associated with preterm birth when the remaining mixture components were held at their median concentrations. Couple-based BKMR models showed a similar relative contribution of paternal (PIP: 61%) and maternal (PIP: 77%) preconception mixtures on preterm birth. We found a positive joint effect on preterm birth across increasing quantiles of couples' total mixture concentrations.ConclusionIn this prospective cohort of subfertile couples, maternal BPA and DEHP, and paternal DEHP exposure before conception were positively associated with preterm birth. Both parental windows jointly contributed to the outcome. These results suggest that preterm birth may be a couple-based pregnancy outcome.

Project description:BackgroundMolar sums are often used to quantify total phthalate exposure, but they do not capture patterns of exposure to multiple phthalates.ObjectiveTo introduce an exposure burden score method for quantifying exposure to phthalate metabolites and examine the association between phthalate burden scores and Homeostatic Model Assessment for Insulin Resistance (HOMA-IR).MethodsWe applied item response theory (IRT) to data from 3474 adults aged 20-60 years in the 2013-2018 National Health and Examination Survey (NHANES) to quantify latent phthalate exposure burden from 12 phthalate metabolites. We compared model fits of three IRT models that used different a priori groupings (general phthalate burden; low molecular weight (LMW) and high molecular weight (HMW) burdens; and LMW, HMW and DEHP burden), and used the best fitting model to estimate phthalate exposure burden scores. Regression models assessed the covariate-adjusted association between phthalate burden scores and HOMA-IR. We compared findings to those using molar sums. In secondary analyses, we examined how the IRT model could be used for data harmonization when a subset of participants are missing some phthalate metabolites, and accounted for measurement error of the phthalate burden scores in estimating associations with HOMA-IR through a resampling approach using plausible value imputation.ResultsA three correlated factors model (LMW, HMW and DEHP burdens) provided the best fit. One interquartile range (IQR) increase in DEHP burden score was associated with 0.094 (95% CI: 0.022, 0.164, p = 0.010) increase in log HOMA-IR, co-adjusted for LMW and HMW burden scores. Findings were consistent when using log molar sums. Associations of phthalate burden and insulin resistance were also consistent when participants were simulated to be missing some phthalate metabolites, and when we accounted for measurement error in estimating burden scores.ConclusionBoth phthalate molar sums and burden scores are sensitive to associations with insulin resistance. Phthalate burden scores may be useful for data harmonization.

Project description:BackgroundPrenatal exposure to multiple phthalates is ubiquitous, and yet few studies have evaluated these exposures as a mixture in relation to child autistic traits and behavioral problems.ObjectivesTo assess cumulative associations between prenatal phthalate mixtures and child behaviors, including effect modification by exposure timing and child sex.MethodsAnalyses included 501 mother/child pairs from the multicenter pregnancy cohort The Infant Development and Environment Study (TIDES). Nine maternal urinary phthalate metabolites were measured in early and late pregnancy, and behavior was assessed at ages 4-5 years using composite T scores for the Behavioral Assessment System for Children (BASC-2), which measures several dimensions of child behavior, and the Social Responsiveness Scale (SRS-2), which measures social impairment consistent with autistic traits. We utilized weighted quantile sum (WQS) regressions to examine pregnancy period-specific associations between phthalate mixtures and behavioral outcomes. Full-sample 95% WQS confidence intervals are known to be anti-conservative, so we calculated a confirmatory p-value using a permutation test. Effect modification by sex was examined with stratified analyses.ResultsA one-quintile increase in the early pregnancy phthalate mixture was associated with increased SRS-2 total score (coefficient = 1.0, confirmatory p = 0.01) and worse adaptive skills (coefficient = -1.0, confirmatory p = 0.06) in both sexes. In sex-stratified analyses, the early pregnancy phthalate mixture was associated with increased SRS-2 total score in boys (coefficient = 1.2, confirmatory p = 0.04) and girls (coefficient = 1.0, confirmatory p = 0.10) and worse BASC-2 adaptive skills score in girls (coefficient = -1.5, confirmatory p = 0.06), while the late pregnancy phthalate mixture was associated with increased BASC-2 externalizing score in boys (coefficient = 1.3, confirmatory p = 0.03).ConclusionOur results suggest cumulative adverse associations between prenatal phthalate mixtures and multiple facets of childhood behavior.

Project description:BackgroundEmerging adults (aged 18-29) report high levels of musculoskeletal pain; however, it is unknown if body location and intensity patterns are related to different biopsychosocial characteristics. This study identified patterns of self-reported musculoskeletal pain among emerging adults and assessed if there were differences in their lifestyle and psychological characteristics.MethodsData from survey responses from a large public university and a large medical university in the United States were used (n = 1,318). Self-reported pain location and intensity at five body regions were assessed, and latent class analysis identified classes separately for men and women. Mental health, physical activity, and sleep outcomes were compared between the classes.ResultsFour classes were identified for men and women. Three of the classes were consistent between genders - "no pain," (women = 28% of their sample; men = 40% of their sample) "mild multisite pain," (women = 50%; men = 39%) and "moderate-severe multisite pain" (women = 9%; men = 7%). The fourth class for women was "moderate spine pain," (13%) and for men was "mild extremity pain" (13%). For both men and women, the "moderate-severe multisite" pain classes reported the highest levels of depression, anxiety, and stress, poorer sleep, and higher work physical activity than the "no pain" class. The "mild multisite" and "moderate spine" (women only) pain classes fell between the "no pain" and "moderate-severe" pain classes. The characteristics of the "mild upper extremity pain" class for men was similar to the "no pain" class.ConclusionsThe identified classes provide unique information on pain location and intensity in emerging adults. The high prevalence of "mild multisite pain" (n = 593; 45% of the total sample) demonstrates an intervention opportunity during this age range to prevent further increases in musculoskeletal pain later in life. Future work should assess the longitudinal outcomes of these pain classes, the impact of interventions for this age group, and the balance between leisure and occupational physical activity when addressing musculoskeletal health.

Dataset Information

Latent classes for chemical mixtures analyses in epidemiology: an example using phthalate and phenol exposure biomarkers in pregnant women.

Publications

Latent classes for chemical mixtures analyses in epidemiology: an example using phthalate and phenol exposure biomarkers in pregnant women.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets